Introduction. The chapter reports four main findings:

Transcription

1

2

3 Developing economies face downside risks to growth and prospects of rising financing costs. In the event that these cause a cyclical slowdown, policymakers may need to employ fiscal policy as a possible tool for stimulus. But will developing economies be able to use fiscal policy effectively? This chapter argues that fiscal space is essential for both the availability and the effectiveness of fiscal policy. Developing economies built fiscal space in the runup to the Great Recession of 28 9, which was then used for stimulus. This reflects a more general trend over the past three decades, where availability of fiscal space has been associated with increasingly countercyclical (or less procyclical) fiscal policy. Wider fiscal space also appears to make fiscal policy more effective. However, fiscal space has shrunk since the Great Recession and has not returned to pre-crisis levels. Thus, developing economies need to rebuild buffers at a pace appropriate to country-specific conditions. For many countries, soft oil prices provide a window of opportunity to implement subsidy reforms that help build fiscal space while, at the same time, removing long-standing distortions. Over the medium-term, credible and well-designed institutional arrangements, such as fiscal rules, stabilization funds, and medium-term expenditure frameworks, can help build fiscal space and strengthen policy outcomes. 1 Introduction Growth in developing economies has slowed in recent years and significant downside risks remain, including slowdowns in major trading partners. In addition, financing costs are expected to rise from the current exceptionally low levels when monetary policy normalization gets under way in some advanced economies. Tightening of global financial conditions and bouts of financial market volatility might cause slowdowns or reversals of capital inflows. 2 Since the risk to capital flows can constrain monetary policy in developing economies, the option of fiscal policy as a countercyclical tool becomes particularly important. 3 How effective will fiscal policy be in supporting activity in developing economies in the event of a downturn? This question is the main focus of the chapter. 1This chapter is prepared by a team led by Ayhan Kose and Franziska Ohnsorge, and including S. Amer Ahmed, Raju Huidrom, Sergio Kurlat, and Jamus J. Lim, with contributions from Israel Osorio- Rodarte and Nao Sugawara, as well as consultancy support from Raphael Espinoza, Ugo Panizza, and Carlos Végh. 2For a discussion on the potential impact of monetary policy normalization on growth and capital inflows in developing economies, see World Bank (214a) and IMF (214a). 3Countercyclicality of fiscal policy refers to an increase in government consumption or cut in taxes during downturns to support economic activity. In the empirical analysis, countercyclicality is defined as a negative and statistically significant response of government consumption to exogenous movements in GDP, as inferred from an econometric model. The chapter also examines countercyclicality in terms of negative and statistically significant correlations between the cyclical components of government consumption and GDP. See Technical Annex for details. There are two related prerequisites for fiscal policy to be useful. First, availability: governments need to have the necessary fiscal space to implement countercyclical measures. Second, effectiveness: countercyclical fiscal policy has to be actually effective in raising the level of economic activity. 4 This chapter draws policy lessons by analyzing the historical experience of developing economies and answering the following questions: How has fiscal space evolved over time? Have developing economies graduated from the procyclicality of fiscal policy during the 198s? Has greater fiscal space supported more effective fiscal policy? What institutional arrangements might strengthen fiscal space and policy outcomes, drawing lessons from country experiences? What objectives with respect to fiscal space should policymakers pursue in the current environment? The focus here is on Emerging Market Economies (EMEs) and Frontier Market Economies (FMEs) that are able to tap international capital markets. 5 The chapter also briefly explores the role of fiscal policy in stimulating activity in Low Income Countries (LICs) that depend on concessional finance. The chapter reports four main findings: During the 2s, in the runup to the Great Recession of 28 9, EMEs and FMEs built fiscal space by reducing debt and closing deficits (Figure 3.1). To support activity during the Great Recession, this space was used for fiscal stimulus. Deficits rose and have remained elevated as EMEs and FMEs have taken advantage of historically low interest rates. Fiscal policy in EMEs and FMEs has become more countercyclical (or less procyclical) since the 198s, as most clearly demonstrated during the Great Recession. Wider fiscal space is associated with more effective fiscal policy in developing economies: fiscal multipliers tend to be larger in countries with greater fiscal space. 4The changing nature of fiscal policy, its availability, and effectiveness in advanced and developing economies have received attention in recent research. Vegh and Vuletin (213) show how fiscal policy has become increasingly countercyclical in Latin America. Ilzetzki et al (213) and Auerbach and Gorodnichenko (212a) explore the effectiveness of fiscal policy in various samples of advanced economies and large emerging markets. Kraay (212) and Eden and Kraay (214) examine the impact of fiscal policy in low-income countries. 5See Annex 3B for details on country classification. 121

4 FIGURE 3.1 Evolution of fiscal space and financing costs Fiscal space used during the crisis has not been rebuilt and EMEs and FMEs are still taking advantage of historically low financing costs to run deficits. A. Fiscal balance Percent of GDP 3 EME FME LIC C. EME long-term interest rates Yield to maturity on 1-year government bond, percent Russian Crisis (October 1998) Dot-Com bubble crash (March 2) Global Crisis (October 28) B. Government debt Percent of GDP D. FME long-term interest rates Estimated yield to maturity on 1-year government bond, percent EME FME LIC Note: All figures are based on unweighted averages across the country grouping or time period. The interest rates over a given time period are averages of daily rates. For EMEs, the nominal long-term interest rate is equal to the government 1-year bond yield. In the case of FMEs, the generic bond yield data were sparse for many economies and time periods. Hence, the nominal interest rate is estimated as the sum of 1-year U.S. Treasury yields plus the predicted spreads from a fixed-effect OLS regression of J.P. Morgan s EMBI on the Institutional Investor Rating. For the crisis periods, the interest rates refer to the average of daily rates in that month. EME: emerging market economies; FME: frontier market economies; LIC: low income countries. Details on the fiscal space data and market based country classifications are described in the Annex 3B. Orange and red bars indicate spikes in longterm interest rates during the relevant months. Well-designed and credible institutional frameworks, such as fiscal rules, stabilization funds, and mediumterm expenditure frameworks, can help build fiscal space and strengthen policy outcomes. In developing economies, debt stocks on average remain moderate despite being higher than expected immediately after the crisis. Fiscal deficits are substantial and have not yet returned to pre-crisis levels. Many economies will need to reduce their fiscal deficits to more sustainable levels. The appropriate speed of adjustment towards these medium-term goals, however, depends on a range of country-specific factors, in particular the cyclical position of the economy and constraints on monetary policy. With restored space, fiscal policy will be more effective in providing support to activity in developing economies than under the current fiscal conditions. The rest of the chapter is organized as follows. The next section describes the conceptual framework for defining and measuring fiscal space. It also outlines the evolution of fiscal space and fiscal policy in EMEs and FMEs. Next, using an econometric model, the chapter estimates fiscal multipliers, which depend on fiscal space. It then discusses institutional arrangements designed to implement sound fiscal policy. The next section assesses current risks, and appropriate medium-term operational goals. The chapter concludes with a brief summary of the main findings and policy recommendations. How Has Fiscal Space Evolved? Definition of Fiscal Space A range of definitions for fiscal space is used in the literature. This chapter follows the definition of Ley (29): availability of budgetary resources for a specific purpose without jeopardizing the sustainability of the government s financial position or the sustainability of the economy. This broad definition allows fiscal space to be considered along multiple dimensions. 6 The first is fiscal solvency risk. The second delineates balance sheet vulnerabilities, such as maturity profile and nonresident shares of government debt, which could generate rollover or liquidity risk for sovereign debt. The third dimension involves factors that could stress private sector balance sheets, and eventually lead to the buildup of contingent fiscal liabilities such as the ratio of external debt-to- GDP or to foreign reserves, the share of short-term debt in external debt, and domestic credit to the private sector relative to gross domestic product (GDP). In line with the literature, this chapter tracks fiscal space mainly in terms of fiscal solvency. Fiscal solvency risk is measured in three alternative ways to capture different elements: first, the government debt-to-gdp ratio (a stock measure of current debt sustainability); second, the fiscal balance-to-gdp ratio (a flow measure of debt accumulation, indicating future debt sustainability, and also one of the measures of rollover risk); and third, the sustainability gap. The sustainability gap is defined as the difference between the actual primary balance and the 6This multidimensional definition helps address the ambiguity of how fiscal space is defined in much of the literature (Perotti, 27). Heller (25) describes fiscal space more broadly as the budgetary room that allows a government to provide financial resources for a specific activity without affecting its financial sustainability while Ostry et al. (21) defines fiscal space specifically as the difference between the current public debt and their estimate of the debt limit implied by the economy s history of fiscal adjustments. 122

5 debt-stabilizing primary balance, which depends on the target debt-to-gdp ratio to be achieved in the long run, the interest rate, and growth. 7 This last measure recognizes that debt sustainability depends on output growth and interest rates, as well as on outstanding debt and deficits. In addition to these measures of fiscal solvency risk, the chapter briefly discusses some aspects of balance-sheet vulnerabilities and private-sector debt. Evolution of Space during the 2s Between 21 and 27, in the runup to the Great Recession, fiscal space widened for much of the developing world, with government debt ratios falling and fiscal deficits closing (Figures 3.1 and 3.2). Three factors contributed to these changes. First, there was rapid growth, with government revenues in commodity exporting economies bolstered by high and rising prices (Figure 3.3). This coincided with a period of increasing graduation of developing economies fiscal policy from earlier procyclicality to more recent countercyclicality. Second, debt relief initiatives, such as the Heavily Indebted Poor Countries (HIPC) Initiative and Multilateral Debt Relief Initiative (MDRI), helped to reduce debt sharply in many FMEs and LICs. 8 As a result, most developing economies consolidated their finances in the early 2s. Third, institutional arrangements in developing economies allowed for improvements in debt management, which also contributed to the reduction in debt-to-gdp ratios (Anderson, Silva and Valendia- Rubiano, 211; Frankel, Vegh, and Vuletin, 213). During the Great Recession, fiscal space narrowed as economies implemented fiscal stimulus. 9 For example, the Republic of Korea boasted wide fiscal space in 27, when government debt was a third of GDP, and fiscal balance was in surplus. In response to the crisis, the government implemented two fiscal stimulus packages, amounting to 7The debt stabilizing primary balance is defined as the primary balance that allows debt to converge to a target debt-to-gdp ratio. This is assumed to be the median stock of public debt as a share of a GDP for a given country grouping. The primary balance is the fiscal balance net of interest expense. Throughout this chapter, government debt refers to gross general government debt unless otherwise specified. See Annex 3B for additional details. 8As of 214, 35 countries have reached the HIPC completion point and are eligible for assistance under the initiative, of which six are FMEs and 22 are LICs (IMF, 214b). The most recent assessment of debt relief costs by the IMF (213) determined that $126 billion has been committed under these initiatives to the 35 HIPC completion point countries, with another $442 million committed to Chad (an interim HIPC country), Cambodia, and Tajikistan. The latter two countries are non-hipc. 9See Eskesen (29), Arbatli et al. (21), and Fardoust, Lin, and Luo (212) for a detailed discussion. 3.6 percent of GDP in 29 and 1.2 percent of GDP in 21. Korea s surplus has diminished since then and debt is now almost 38 percent of GDP. Similarly, China had a fiscal surplus in 27, and government debt that was just one-fifth of GDP. Following a stimulus package equivalent to 12.5 percent of GDP in 28, China ran fiscal deficits from 28 to 21. Government debt rose to more than 5 percent of GDP by Both economies succeeded in preventing a contraction in real GDP, despite the sharp downturn in the global economy. Space and Policy during Contractions China and Korea were particularly pronounced examples of a broader pattern among EMEs and FMEs. Many implemented countercyclical fiscal policy during the Great Recession, but not all avoided GDP contractions. To analyze fiscal policy responses during the Great Recession as well as in past crises, the chapter conducts an event study that identifies 11 episodes of sharp annual GDP contractions in 157 advanced and developing economies since 199 (see Annex 3A for details). A country is considered to have experienced a contraction event if its GDP growth in a given year fulfills two conditions: first, growth is negative (i.e., a contraction), and second, growth is more than one standard deviation below the average growth that the economy experienced over These criteria yielded 51 economies in the sample that experienced a contraction during the Great Recession, of which 21 were EMEs or FMEs. 11 During the Great Recession, EMEs and FMEs used the wider fiscal space they had accumulated during the preceding years to allow automatic stabilizers to operate and to implement larger fiscal stimulus than in earlier contractions. Structural balances, which measure the fiscal policy stance, declined sharply as economies entered severe contractions (Figure 3.4). 12 During both event samples, fiscal space deteriorated following the stimulus, reflected in an increase in government debt. Government debt evolved differently across the two samples, likely as a result of different exchange rate movements and financial sector support programs. 1The buildup of general government debt reflected a substantial expansion in local government off-balance sheet lending (World Bank, 213a, 214b). 11More than 8 percent of advanced market countries (AMEs), a third of EMEs and FMEs, and less than a tenth of LICs experienced a contraction in 28-9 in the sample of countries considered. 12In this chapter, the structural balance is defined as the difference between cyclically-adjusted revenues and cyclically-adjusted expenditures. It thus removes the cycle-induced component of taxes and expenditures, such as social safety nets. See Statistical Annex for additional details. 123

6 FIGURE 3.2 Government debt in 21 and 27 The combination of strong growth, high commodity prices, and debt relief initiatives helped developing economies gain fiscal space in the runup to the Great Recession. A. 21 B. 27 Note: A greener color indicates lower government debt as a percentage of GDP and a redder color indicates higher government debt as a percentage of GDP. 124

7 FIGURE 3.3 Fiscal space in commodity exporters and importers The gains in fiscal space were more pronounced for commodity exporters. FIGURE 3.4 Fiscal policy and space during contractions EMEs and FMEs made greater use of fiscal stimulus during the Great Recession than during earlier contractions. A. Government debt A. Structural balance B. Government debt Percent of GDP Commodity importers Commodity exporters Percent of potential GDP All events pre-28 Great Recession Percent of GDP All events pre-28 Great Recession C. Exchange rate index D. GDP growth B. Fiscal balance Percent of GDP 6 Commodity importers Commodity exporters 4 Index is 1 at t= All events pre-28 Great Recession Percent All events pre-28 Great Recession World average at t= during events pre-28 World average at t= during Great Recession Note: t= is the year of the trough of the contraction episode. All variables refer to the unweighted sample mean. The structural balance is defined as the difference between cyclically adjusted revenues and cyclically adjusted expenditures. The exchange rate index is set to be 1 at t= and shows how exchange rates depreciated in pre-28 contraction episodes but not during the Great Recession. The world average growth during pre-28 contraction episodes was much higher than during the Great Recession, and so economies experiencing contractions in 28-9 did so under more difficult global conditions than in previous contractions Note: Commodity exporters include all oil and mineral exporting economies that are identified as such by the Global Economic Prospects. Commodity importers are all economies that are not classified as exporters. Figures refer to unweighted averages of commodity importers and exporters data. In particular, in pre-28 contractions, sharp exchange rate depreciations raised the cost of holding foreign currency debt and contributed to steep increases in the debt ratio. Cases in point are the Asian crisis and the Russian crisis of the late 199s. 13 In comparison, during 28 9, EME and FME currencies dropped less and rebounded to pre-crisis levels before the Great Recession 13Kohler (21) documents the differences in exchange rate depreciations between the 28 9 crisis and the Asian and Russian crises. Didier, Hevia, and Schmukler (212) show that there were structural breaks in policy in EMEs, based on a comparison between policies in the Asian and Russian crises and the Great Recession. EMEs experienced smaller depreciations during the Great Recession. Moreover, EMEs lost substantially less reserves during the crisis than during the Asian and Russian crises. was over. This partly reflected a different, more difficult, global environment with a somewhat deeper contraction and weaker global recovery. The risks posed by exchange rate depreciation may be smaller for emerging economies now than in the past, due to deeper domestic financial markets and a policy decision to borrow in domestic currency, thus reducing original sin. 14 In addition, before 28, some EMEs suffered systemic banking crises which required governments to provide heavy financial support. Though typically not fully reflected in deficits, such outlays substantially increased public debt above and beyond the increases attributable to the fiscal deficit (Laeven and Valencia, 213). As these cross-country experiences illustrate, the fiscal space implicit in low debt can shrink rapidly especially during periods of elevated financial stress (Figure 3.5). 14Original sin refers to the inability of some developing countries to borrow internationally in their own currency (Eichengreen and Hausmann, 1999). Hausmann and Panizza (211) analyze the risks posed by original sin. 125

8 GLOBAL ECONOMIC PROSPECTS January 215 Chapter 3 FIGURE 3.5 Government debt in select crises Debt can rise very quickly during a crisis episode. Percent of GDP Indonesia Thailand Ireland Latvia Note: Central government debt is used for Indonesia. The others refer to general government debt. Have Developing Economies Graduated from Procyclicality? There are several measures of the stance of fiscal policy. This chapter employs two that are commonly used in the literature: the structural balance and government consumption. The structural balance strips from the overall balance the rise and fall of revenues (such as the cycle-induced component of income taxes) and expenditures (especially social benefits) that can be attributed to the business cycle. The other measure, government consumption expenditures, which are mainly government wages and outlays on goods and services, provides a narrower definition of the fiscal policy stance, but one that is more readily comparable across economies and not subject to the uncertainty surrounding the accuracy of cyclical adjustments, for example the uncertainty about the cyclical income elasticity of tax revenues or the size of the output gap. On either measure, fiscal policy was significantly more expansionary during the Great Recession than during earlier contraction episodes. Structural balances widened, on average among EMEs and FMEs, by 4 percentage points of GDP during the Great Recession, whereas they tightened in earlier contractions. The buildup of fiscal space during the global expansion of the early 2s, and its use during the Great Recession suggest that fiscal policy has become less procyclical in developing economies. Estimated responses of government consumption to GDP shocks indeed show that fiscal policy has become less procyclical since the 199s, and more countercyclical since the Great Recession (Figure 3.6). 15 While the sample is too small to compute estimates for EMEs and FMEs separately, correlations between real GDP and real government consumption also suggest a similarity between the two groups. High procyclicality between 198 and 1999, broadly turned to acyclicality in EMEs in the early 2s, and to countercyclicality after the Great Recession. This evolution of fiscal cyclicality can be attributed to several factors, including improvements in policies, institutions, and enhanced financial market access. 16 The move to less procyclical fiscal policy has also been associated with greater fiscal space. Throughout the 2s, procyclicality was less pronounced in economies with wide fiscal space (Figure 3.7). During the Great Recession, economies with government debt below 4 percent of GDP (implying wider fiscal space) were able to implement greater fiscal stimulus than more indebted governments (with narrower space) (Figure 3.8). Fiscal policy in LICs has remained mostly acyclical reflecting the severe budgetary constraints they often face (Box 3.1). 17 Overall, the evidence presented in this section suggests that fiscal space matters for a government s ability to implement countercyclical fiscal policy. The next section explores the importance of space for policy effectiveness. Does Greater Space Tend to Support More Effective Fiscal Outcomes? Countries with more ample fiscal space have used stimulus more extensively during the Great Recession than those with tighter space. But has this stimulus been more effective at meeting the goal of supporting activity? Space may affect the effectiveness of fiscal policy through two channels. Interest rate channel: When fiscal space is narrow, expansionary policy can increase lenders perceptions 15These responses are estimated using a vector autoregressive model (VAR) with a pooled sample of EMEs and FMEs. See Technical Annex for details of the VAR model. 16Frankel, Végh, and Vuletin (213) emphasize the importance of improvements in institutional quality for the changes in cyclicality. Calderon and Schmidt-Hebbel (28) and World Bank (213b) discuss the importance of greater credibility of fiscal policies and deepening domestic financial markets. 17World Bank (213b) offers explanations of the procyclical bias of fiscal policy in developing countries. Developing countries have generally procyclical access to capital markets, and governments must therefore make spending cuts during downturns, when they are less able or unable to borrow. During upswings, governments are often under political pressure to spend the higher revenues. 126

9 GLOBAL ECONOMIC PROSPECTS January 215 Chapter 3 FIGURE 3.6 Changing stance of fiscal policy Fiscal policy has become countercyclical (or less procyclical) in EMEs and FMEs since the 198s. A. Impulse responses of government consumption to GDP shocks B. Correlations between government consumption and GDP EME FME Note: The cumulative impulse responses of government consumption (in percent) at the one-year horizon following a 1 percent positive shock to GDP. The impulse responses are estimated using a panel SVAR model with a sample of 15 EMEs and FMEs (see Annex 3A for details of the model and Table 3B.2 in Annex 3B for the list of countries). Note: Presents correlations between the cyclical components of government consumption and GDP from an unbalanced panel of annual data for 31 EMEs and 29 FMEs. All correlations are statistically significantly different from zero and differences in correlations across time are also statistically significant. Positive responses (Panel A) and positive correlations (Panel B) suggest procyclicality, while negative responses (Panel A) and negative correlations (Panel B) suggest countercyclicality. FIGURE 3.7 Cyclicality of fiscal policy and fiscal space FIGURE 3.8 Structural balance during the Great Recession In the 2s, fiscal policy was countercyclical (or less procyclical) in countries with wider fiscal space. Countries with wider fiscal space implemented larger stimulus packages during the Great Recession. Correlation between government consumption and GDP Percent of potential GDP Narrow fiscal space EME Wide fiscal space Narrow fiscal space FME Wide fiscal space Wide fiscal space Narrow fiscal space Note: The correlations are between the cyclical components of government consumption and GDP with samples divided based on fiscal space from an unbalanced panel of annual data for 31 EMEs and 29 FMEs. The median debtto-gdp ratio in the full sample is 44 percent. Countries with debt-to-gdp ratios above the median are considered to have narrow fiscal space, while those with debt-to-gdp ratios below the median are considered to have wide fiscal space. All correlations are statistically significantly different from zero and across time. Positive correlations suggest procyclicality, while negative correlations suggest countercyclicality. Note: t= is the year of the trough of the contraction episode. All variables refer to the unweighted sample mean. These results are based on the data sample of the event study which includes the 21 EMEs and FMEs that experienced contractions during the Great Recession. The median debt-to-gdp ratio in the full sample of 63 EMEs and FMEs is 44 percent. Countries with debt-to-gdp ratios above the median are considered to have narrow fiscal space, while those with debt-to-gdp ratios below the median are considered to have wide fiscal space. 127

10 BOX 3.1 Fiscal Policy in Low-Income Countries 1 Fiscal policy in low-income countries (LICs) has been largely acyclical over the past two decades as shown by very low correlations between the cyclical components of government consumption and GDP during this period (Figure B3.1.1, panel A). This suggests that LICs do not systematically use fiscal policy to stabilize the business cycle. But when they do, how effective is fiscal policy? Empirical estimates of the multipliers in LICs are few, partly because the identification of an exogenous fiscal shock imposes stringent data requirements. One approach, used in Kraay (212, 214), is to identify a fiscal shock using World Bank loan disbursements. First, loans disbursed by the World Bank are a major source of finance for government spending in LICs. Second, the timings of approval and disbursement of such loans are not systematically related to cyclical macroeconomic conditions in recipient countries. This makes World Bank loans a good instrument for exogenous government spending, unrelated to cyclical macroeconomic conditions in LICs. Using this approach, the average (one-year) fiscal multipliers in LICs are estimated to be small at about.5. The second approach is to apply a panel structural vector auto regression (SVAR) model to annual data the only frequency available for LICs on a comparable cross-country basis for government consumption and GDP. A fiscal shock is identified by a similar timing assumption used in Blanchard and Perotti (22) except that now it is assumed that discretionary fiscal policy takes at least a year (and not a quarter) to respond to macroeconomic conditions. Such a prolonged lag in the response of discretionary fiscal policy may be justified in LICs on two grounds. First, LICs often rely on concessional loans to finance government spending and these are disbursed less frequently than every quarter and may discount macroeconomic conditions. Second, GDP data is extensively revised in these economies so that the government would likely take more than just one quarter to gather reliable GDP data (Ley and Misch, 214). This then implies that discretionary fiscal policy aimed at stabilizing the economy would take more than just one quarter to implement. Fiscal multipliers are estimated using annual data for 34 low income-economies and a panel SVAR following the methodology of Ilzetzki, Mendoza, and Vegh (213). The multiplier estimates are just above.6 (Figure B3.1.1, panel B), closely in line with the results from Kraay (212, 214). Government financing in LICs is mostly concessional and not market based. Hence, market concerns about government solvency that underpin the relationship between fiscal space and FIGURE B3.1.1 Note: Panel A shows the correlation between the cyclical components of government consumption and GDP. The correlations are all statistically insignificant which suggest that fiscal policy is acyclical in LICs. Panel B shows the fiscal multipliers based on a panel SVAR model. See Annex 3A for the details. 1The main author of this box is Raju Huidrom. Cyclicality and multipliers in LICs Fiscal policy is acyclical in LICs and multipliers are relatively small. A. Cyclicality of fiscal policy B. Fiscal multipliers Impact 1 Year 2 years multipliers are expected to be less relevant in LICs than in EMEs and FMEs. Therefore, fiscal multipliers likely do not vary significantly with fiscal space in LICs. That said, fiscal space remains important in LICs, because it ensures that countercyclical fiscal policy is available when needed. 128

11 of sovereign credit risk. This raises sovereign bond yields and hence, borrowing costs across the whole economy (Corsetti et al., 213; Bi, Shen, and Yang, 214). This, in turn, crowds out private investment and consumption. If the crowding out is sufficiently strong, the net effect of expansionary fiscal policy on output, that is, the size of the fiscal multiplier, may be negligible or even negative. Ricardian channel: When a government with narrow fiscal space conducts a fiscal expansion, households expect tax increases sooner than in an economy with wide fiscal space (Perotti, 1999; Sutherland, 1997). The perceived negative wealth effect encourages households to cut consumption and save, thereby weakening the impact of the policy on output. 18 The effectiveness of fiscal policy is usually evaluated in terms of the fiscal multiplier the change in output for a dollar increase in government consumption. The more positive the multiplier, the more effective is policy. For developing economies, the literature reports multipliers that are small in size, and variable, ranging from -.4 to.9 (Box 3.2). These estimates often refer to average multipliers, over a whole range of macroeconomic conditions. Recent work in the context of advanced economies has found that multipliers vary significantly depending on macroeconomic conditions and country characteristics: they tend to be larger during recessions (Auerbach and Gorodnichenko, 212a, 212b), for economies using a fixed exchange rate regime, and for economies with low debt (Ilzetzki, Mendoza, and Vegh, 213, based on pre-crisis data; Nickel and Tudyka, 213, for OECD economies). To estimate fiscal multipliers for developing economies that depend on fiscal space, this section employs an Interacted Panel VAR (IPVAR) model (Towbin and Weber, 213). This allows model parameters, and hence estimated fiscal multipliers, to interact with fiscal space. Fiscal shocks are identified by assuming that discretionary policy takes at least one quarter to respond to macroeconomic conditions (Blanchard and Perotti, 22). The variables included in the model are government consumption, GDP, current account balance, and real 18While crowding-out effects of fiscal policy, that operate via higher interest rates or future increase in taxes, have long been discussed in the literature, the emphasis in this chapter is that such crowding-out effects can be nonlinear and can depend on fiscal space. In particular, the nonlinearity pertains to investors perception of sovereign credit risk (interest rate channel) and households expectation of future tax increases as fiscal space becomes narrow (Ricardian channel). The interest rate channel is less relevant for large advanced economies that are able to issue debt in their own currency (Krugman, 211). effective exchange rates. 19 The baseline results are based on an unbalanced panel for 15 EMEs and FMEs (augmented by 19 advanced economies in robustness exercises). The data are quarterly, 198:1 214:1. Fiscal policy is proxied by government consumption. 2 The model estimates fiscal multipliers as a function of fiscal space, which is proxied by fiscal balances as percent of GDP, corresponding to a flow measure. To control for endogeneity and to ensure that fiscal balances do not systematically pick up business cycle effects, lagged moving averages of fiscal balances are employed. 21 The results (Figure 3.9) suggest that the multipliers at the one-year horizon are not much above zero when preexisting fiscal deficits leading up to the stimulus have been high (narrow fiscal space), but are positive and significant when there have been surpluses (wide fiscal space). 22 The multipliers at the two-year horizon are generally greater than at the one-year horizon, suggesting that the effects peak with some lag. At longer horizons, multipliers remain near zero and statistically insignificant when fiscal space is narrow, but can be as high as 1.8 when fiscal space is wide. This result is qualitatively robust to alternative measures of fiscal space. For example, the results for the multipliers that use the sustainability gap as the gauge of fiscal space also point to these conclusions (Figure 3.1). The results are similar when government debt as percent of GDP is used as the measure of fiscal space (see Annex 3A). In addition to the baseline model above, two alternative econometric models are used to examine robustness: a panel Structural VAR (SVAR) as in Ilzetzki, Mendoza, and Vegh (213), and a local projections model as in Riera-Crichton, Vegh, 19This follows Ilzetzki, Mendoza, and Vegh (213). 2Since data availability and comparability is limited for the EMEs and FMEs included here, the analysis does not address the issue of spending composition, although this may be important. For instance, government spending on infrastructure and health has been shown to protect and strengthen social safety net programs, and result in long-run growth benefits (Berg et al., 29; Kraay and Serven, 213). Public infrastructure investment multipliers are often much larger than the public consumption multipliers (IMF, 214c). The analysis here also does not cover automatic stabilizers which, at least in the case of OECD countries, has played a strong role in stabilizing output (Fatás and Mihov, 212). 21Indeed, this fiscal space measure is not systematically wider during recessions than expansions in the sample of EMEs and FMEs. For example, the average fiscal deficit during recessions is 2.7 percent of GDP, which is very close to the deficits during expansions, 2.8 percent of GDP. Alternatively, the regression coefficients could be interacted with an additional dummy for recessions. However, this reduces the degrees of freedom significantly and results in imprecise estimates. The fiscal space measure also does not reflect exchange rate regimes the proportion of fixed and flexible exchange rate regimes in the sample is roughly the same during periods of wide and narrow fiscal space. 22The multipliers presented here are the cumulative multipliers that take into account the persistence in the response of government consumption due to a fiscal shock. See Annex 3A for details. 129

12 BOX 3.2 What Affects the Size of Fiscal Multipliers? 1 The size of fiscal multipliers depends on macroeconomic conditions and country-specific features. While the chapter examines how fiscal multipliers depend on fiscal space, especially in the context of developing economies, this box reviews additional aspects that have been important in explaining the size of multipliers. Conditions affecting multipliers Fiscal multipliers depend on the phase of the business cycle: they tend to be larger during recessions than during expansions (Auerbach and Gorodnichenko, 212a, 212b). In theory, this is attributed to a higher level of economic slack (Rendahl, 212) and a greater share of liquidity-constrained households (Canzoneri et al., 212) during economic downturns. The effectiveness of fiscal policy also depends on monetary policy. Monetary contraction, in response to expansionary fiscal policy that increases inflation and output, blunts the effects of the fiscal policy on output. Similarly, the effects of fiscal policy on output are more pronounced when monetary policy is more accommodative, especially when interest rates are at the zero lower bound (Christiano, Eichenbaum, and Rebelo, 211). The effectiveness of fiscal policy also depends on countryspecific features. Fiscal multipliers tend to be larger in economies with fixed exchange rates than in economies with flexible exchange rates (Ilzetzki, Mendoza, and Vegh, 213) because, in fixed regimes, expansionary fiscal policy tends to trigger some monetary accommodation. Fiscal multipliers are also larger in less open economies because of lower leakages into import demand. Finally, the choice of the fiscal instrument matters. Revenuebased fiscal multipliers tend to be lower (especially in the short term) than expenditure-based multipliers. Expenditures tend to affect aggregate demand directly, whereas changes in revenues operate only indirectly and are subject to leakage. For example, households may save a portion of tax cuts intended to stimulate aggregate demand. Some caution is warranted here as recent work has shown that cyclically adjusted tax revenues are not a good proxy for tax policy. Riera-Crichton, Vegh and Vuletin (212) argue that using tax rates instead of tax revenues yields considerably higher tax multipliers. Empirical estimates Empirical estimation of fiscal multipliers requires a strategy to identify exogenous fiscal shocks. The one deployed in the chapter relies on a timing assumption that discretionary fiscal policy takes at least a quarter to respond to macroeconomic conditions (Blanchard and Perotti, 22). There are alternative identification strategies deployed in the literature: the narrative approach as in Ramey and Shapiro (1998) or Guajardo, Leigh, and Pescatori (214); forecast errors as in Blanchard and Leigh (213); or fluctuations in aid-related financing approval used as instruments in Kraay (212, 214). Fiscal multipliers can also be obtained from estimated dynamic stochastic general equilibrium (DSGE) models (Coenen et al., 212). While empirical approaches yield reduced-form estimates of fiscal multipliers, DSGE-based estimates can capture deep structural features of the economy, in particular the interactions between private-sector behavior and policy parameters. The vast majority of the estimates fall between zero and unity. Multipliers, on average, tend to be somewhat larger in advanced economies relative to developing ones. Recent work, although mostly in the context of advanced economies, has shown that multipliers depend on macroeconomic conditions consistent with the theoretical predictions above. For instance, the size of multipliers tends to be significantly larger during recessions. Estimates place the long-term fiscal multiplier during recessions between.6 and 2.7, which is generally several times larger than multipliers during more tranquil economic conditions. These effects are even larger when interest rates are at the zero lower bound. In addition to the phase of the business cycle, evidence for advanced economies suggests that fiscal multipliers are smaller in the presence of narrow fiscal space, and can even turn negative (Table B3.2.1). 1The main author of this box is Jamus J. Lim. 2Using tax revenues as the fiscal instrument first involves adjusting for the cyclical or the automatic stabilizer component via elasticity estimates. One reason the chapter does not discuss revenue-based multipliers is that elasticity estimates tend to be unreliable for EMEs and FMEs. 13

13 BOX 3.2 (continued) FIGURE 3.9 Fiscal multipliers by fiscal space Fiscal policy in EMEs and FMEs tends to be more effective when fiscal space is wider. A. 1 year TABLE B3.2.1 Gro up s/ F eat ures Fiscal multipliers: A review of studies Sho rt - t erm mult ip lier Lo ng - t erm mult ip lier 3 2 Income group 1 Advanced economies Developing economies Upper-middle income Fiscal balances as percent of GDP Lower-middle income B. 2 years 3 Low income Business cycle 2 Wider fiscal space Expansion Recession Zero lower bound Fiscal space Wide space Fiscal balances as percent of GDP Narrow space Sources: World Bank compilation; Batini et al., (214); Ilzetzki, Mendoza, and Vegh (213); Mineshima, Poplawski-Ribeiro, and Weber (214); and Ramey (211). Notes: Estimates are for both government consumption and expenditure multipliers. Minimum and maximum estimates may refer to distinct studies and/or economies. Where available, short-term multipliers report the impact multiplier; otherwise the multiplier at the one-year horizon is used. Where available, long-term multipliers report the cumulative multiplier at the horizon of five years; otherwise the longest (generally three-year) horizon is used. The highincome and developing multipliers report linear estimates without state dependency. 1 The upper-middle income estimates are skewed by the unusually large multiplier of China (2.8). Hence, China was excluded from the computation of the upper bound. 2 Applies to zero lower bound for monetary policy rates. Multipliers depend heavily on the duration of the period in which the zero lower bound is binding; short-term (long-term) estimates reported here correspond to a zero lower bound of one (twelve) quarters. 3 Fiscal space in these studies is usually measured in terms of the debt-to-gdp ratio: a high (low) debt-gdp ratio indicates fiscal space is narrow (wide). Note: The graphs show fiscal multipliers for different levels of fiscal space at horizons of one and two years. These are based on the estimates from the IPVAR model using a sample of 15 EMEs and FMEs. Fiscal balance as a percentage of GDP is the measure of fiscal space and the values shown on the x-axis correspond to the percentiles from the sample. Fiscal space is narrow (wide) when fiscal balances are low (high). Solid lines represent the median, and shaded areas around the solid lines are the percent confidence bands. and Vuletin (214). 23 Although the precise estimates of the multipliers differ, the results from the alternative models also suggest that fiscal policy is more effective fiscal multipliers are higher when pre-existing fiscal space leading up to the stimulus is wide than when it is narrow (see Annex 3A). In sum, the empirical evidence presented here suggests that wider fiscal space is associated with more effective fiscal policy in developing economies. This result holds for different types of fiscal space measures using various empirical approaches. 23Details of these two models are provided in Annex 3A. 131

14 FIGURE 3.1 A. 1 year Fiscal multipliers and sustainability gap mitigate these pressures and support fiscal discipline. In particular, it highlights best practices for three institutional mechanisms fiscal rules, stabilization funds, and medium-term expenditure frameworks (MTEFs) along with empirical evidence on the relative success of these institutions in strengthening fiscal space and supporting countercyclical fiscal policy. 25 Fiscal Rules B. 2 years Sustainability gap Wider fiscal space Sustainability gap Note: The graphs show fiscal multipliers for different levels of fiscal space at horizons of one and two years. These are based on the estimates from the IPVAR model using a sample of 15 EMEs and FMEs. The sustainability gap, the difference between the actual primary balance and the debt-stabilizing primary balance at current interest and growth rates, is the measure of fiscal space. The values shown on the x-axis correspond to the percentiles from the sample. Fiscal space is narrow (wide) when the sustainability gap is low (high). Solid lines represent the median, and shaded areas around the solid lines are the percent confidence bands. Institutional Arrangements: How Can Fiscal Space Be Strengthened? Fiscal rules impose lasting numerical constraints on budgetary aggregates debt, overall balance, expenditures, or revenues. Rules often allow for flexibility in meeting budget targets by taking into account temporary cyclical deviations such as a large output gap or structural adjustments, such as changes in the medium-term price of a key export. Fiscal rules, and in particular cyclically-adjusted or structural balance rules, have become increasingly popular in developing economies (Figure 3.11), especially since the Great Recession (Schaechter et al., 212). Balanced budget rules have become common in Africa and Eastern Europe, often adopted alongside debt rules. The adoption of rules, per se, has had mixed success in limiting procyclicality. Indeed, balanced budget rules that target headline fiscal balances can lead not only to more volatile business cycles but they also tend to be associated with more procyclical fiscal stances (Bova, Carcenac, and Guerguil, 214). In contrast, budget balance rules that target structural balances tend to be more closely associated with countercyclical fiscal stances. Many countries with budget rules have been transitioning to targeting cyclically-adjusted balance. Other possible factors that explain the limited success of balanced budget rules to reduce procyclicality include challenges to enforcement such as the off-budget government guarantees (World Bank, 214b), insufficient flexibility (Snudden, 213), and the need for greater The past procyclicality of fiscal policy in developing economies has been attributed in part to political economy pressures. 24 This section discusses how credible and well-designed institutional mechanisms can help 24See World Bank (213a) for a more detailed discussion. Volatile foreign capital market access is another constraint discussed in the literature (Cuadra, Sanchez, and Sapriza, 21). 25Broadly speaking, the design of an effective budgeting process that ensures that macroeconomic fiscal targets are met depends on the type of governing approaches. A delegation approach, based on clear hierarchical layers between decision makers, tends to be more effective in countries where governments are formed by a single party, or the electoral process is not competitive. A contract approach, based on agreement between decision makers along largely horizontal relationships, tends to be more effective in countries where coalition governments are the norm, and elections are competitive (Buttiglione et al., 214). Within these two broad approaches, fiscal rules, stabilization funds, and medium-term budgeting frameworks can appropriately constrain discretion, and ensure that budgets are in line with longer-run macroeconomic goals. 132

15 FIGURE 3.11 Fiscal rules: Trends and distribution FIGURE 3.12 Stabilization funds: Trends and distribution A. Trends, A. Trends, Number of countries with balance budget rules Number of funds Dollars per barrel 4 Developing economies Advanced economies 3 Number of stabilization funds Number of oil-related funds Oil price, US$ per barrel (right axis) B. Distribution across developing economies, 213 Number of rules Expenditure rule Revenue rule Balanced budget rule Debt rule B. Distribution across developing economies, 213 Number of funds Pre Post AFR EAP ECA LAC SAR Oil Other Oil Other Oil Other Oil Other Oil Other EAP ECA LAC MNA AFR Notes: The database includes 87 economies. AFR: Sub-Saharan Africa; EAP: East Asia and Pacific; ECA: Europe and Central Asia; LAC: Latin America and Caribbean; SAR: South Asia. There is no reported fiscal rule for the Middle East and North Africa. Notes: Stabilization funds here are all those listed in Sugawara (214), together with Panama s fund (established in 212), but excluding Norway. Oil-related stabilization funds are those whose funding sources include petroleum, the rest are referred to as Other in the graph. Only the first fund each country created is included if multiple funds exist (or existed) in a country. AFR: Sub-Saharan Africa; EAP: East Asia and Pacific; ECA: Europe and Central Asia; LAC: Latin America and Caribbean; MNA: Middle East and North Africa. transparency and improved measurement in the estimation of structural balances. Rules are best when simply defined and supported by surveillance arrangements, respected by the government, yet operated by a non-government agency (Frankel, 211). Chile s use of a technical fiscal council and a fiscal rule that targets a fixed structural balance is a good example of a welldesigned, credible, and successfully operated fiscal rule (Box 3.3). Such agencies have legal guarantees for independence, highly qualified professional staff, and assured financing (Debrun and Schaechter, 214). Stabilization Funds Stabilization funds set aside receipts from significant natural resource revenues such as oil and natural gas. Funds saved during favorable times are released to cushion potential revenue shortfalls and to mitigate negative shocks to government expenditure. Stabilization funds were first set up in Kuwait in 1953, and were adopted widely in the 2s, when high international oil prices along with the discovery of oil in a number of economies facilitated their establishment (Figure 3.12). Many stabilization funds are integrated with the budget, with clear rules to guide the accumulation and withdrawal of fund resources (Bagnall and Truman, 213). 26 Since 26For example, Trinidad and Tobago s Heritage and Stabilization Fund requires that at least 6 percent of total excess petroleum revenues must be deposited into the stabilization fund. Similarly, Timor-Leste s Petroleum Fund Law of 25 requires all receipts from petroleumrelated activities to be transferred to its stabilization fund. 133

16 BOX 3.3 Chile s Fiscal Rule An Example of Success 1 Political pressures that underlie procyclicality of fiscal policy can be partly mitigated by the design of mechanisms (such as fiscal rules or stabilization funds) that are supported by technically sound and credible institutions (such as fiscal councils) (World Bank, 213c). Chile presents an example of a well-designed mechanism in an enabling institutional environment. Chile is the world s largest exporter of copper. It has experienced significant macroeconomic volatility for much of its history due to terms-of-trade shocks associated with fluctuations in global copper prices. In 21, Chile adopted a fiscal regime that was designed to break this pattern. The regime was based on a target for the structurally-adjusted fiscal balance, which adjusted the overall balance for the output gap and commodity prices. Importantly, the determination of both the output gap and the medium-term price of copper is entrusted to two expert panels, comprising representatives from both the private sector and academia, which serve the crucial role of providing unbiased projections of these key variables (Frankel, 211). The role of the government is limited to adjusting expenditures to meet the structural balance target. The Fiscal Responsibility Law that Chile enacted in 26 provides an institutional framework that strengthens the link between the fiscal rule, government savings, and two sovereign wealth funds the Pension Reserve Fund and the Economic and Social Stabilization Fund (Schmidt- Hebbel, 212a; 212b). The law also facilitates greater transparency and disclosure in the conduct of fiscal policy. The introduction of the fiscal regime coincided with a global copper boom, which led to steadily increasing fiscal surpluses, peaking at 7.4 percent of GDP on the eve of the global crisis (Figure B3.3.1). By the end of 27, the government debt-to- GDP ratio had fallen to single digits. As surpluses rose, the council of technical experts stood firm against political pressures to assume that copper prices would remain permanently high and to maintain higher spending levels. Copper prices fell sharply during the Great Recession. The significant fiscal space built up over the preceding years allowed Chile to implement a stimulus package amounting to 2.9 percent of GDP. It included increases in public investment; temporary reductions in a range of taxes; and subsidies for housing, transportation, and lowincome households (IMF, 29). In part because of this fiscal stimulus, growth resumed the following year. While the recovery of the global economy was also accompanied by a rebound in copper prices, they did not return to pre-crisis levels. Chile s fiscal rule and its use of fiscal policy during the crisis illustrate an important limitation of the rule. Chile s rule specifically calls for a zero structural balance, and thus does not FIGURE B3.3.1 A. Fiscal balance, Percent Revenue/GDP Percent, percent of GDP Chile s fiscal indicators and economic performance Expenditure/GDP GDP growth allow the implementation of countercyclical fiscal stimulus. The stimulus of 29 was only implemented with a change in the rule after much deliberation by country authorities. Escape clauses in fiscal rules that accommodate such circumstances can thus provide valuable flexibility in dealing with low probability events and are included in recent fiscal rules (Schaechter et al., 212). 1The main author for this box is Jamus J. Lim. Fiscal balance (RHS) B. Structural fiscal balance, government debt, and growth Debt (RHS) Structural balance Percent of GDP Percent of GDP Notes: Fiscal and structural balance data are from the database described in Annex 3B. GDP data are from the World Development Indicators

17 stabilization funds separate government expenditure from fluctuations in the availability of revenues, they can be important institutional mechanisms for improving fiscal space, while mitigating fiscal procyclicality. Although the empirical evidence is somewhat mixed, a number of studies find that stabilization funds can help improve fiscal discipline (Fasano, 2) and expand fiscal space (Bagattini, 211). Stabilization funds do appear to smooth government expenditure, reducing their volatility by as much as 13 percent compared to economies without such funds (Sugawara, 214). While a stabilization fund can be a powerful fiscal tool to manage fiscal resources and create fiscal space, the establishment itself does not guarantee its success. Crosscountry evidence even suggests that the effectiveness of a particular stabilization fund in shielding the domestic economy from commodity price volatility depends largely on government commitment to fiscal discipline and macroeconomic management, rather than on just the existence of the instrument itself (Gill et al., 214). Proper designs and strong institutional environments that support their operations are crucial factors for the success of stabilization funds. Among resource-rich economies, Norway and Chile are often treated as examples of economies with stabilization funds that are based on specific resource revenues and associated with good fiscal management (Schmidt- Hebbel, 212a, 212b). Norway s Government Pension Fund and Chile s Economic and Social Stabilization Fund are ranked highest and third, respectively, in a scoring of 58 sovereign wealth funds and government pension funds (Bagnall and Truman, 213). The main characteristics that distinguish Norway s and Chile s funds from those with lower scores are governance and transparency and accountability of fund operations. Medium-Term Expenditure Frameworks (MTEF) MTEFs were first introduced to facilitate modern public financial management in pursuit of long-run policy priorities in OECD economies. Among developing economies, they gained prominence in the late 199s, as annual budgets were perceived to create uncertainty about future budgetary commitments. International financial agencies, such as the World Bank, have also sought to encourage stable allocations toward poverty reduction targets. More than two-thirds of all economies have adopted MTEFs of some form (World Bank, 213c). The objective of MTEFs is to establish or improve credibility in the budgetary process. They seek to ensure a transparent budgetary process, where government agencies establish credible contracts for the allocation of public resources toward agreed strategic priorities, over an average of three years. The most common design of MTEFs translates macroeconomic objectives into budget aggregates and detailed spending plans; less sophisticated approaches target either aggregate fiscal goals, or microlevel costs and outcomes. Empirical evidence suggests that credible MTEFs can significantly improve fiscal discipline (World Bank, 213c). Furthermore, the results tend to be more positive for more sophisticated frameworks (Grigoli et al., 212). Significant heterogeneity exists, however, and certain studies limited to smaller regional samples have been unable to find conclusive evidence, possibly reflecting shortcomings in the practical implementation of MTEFs. 27 Keys to robust implementation are coordination with broader public sector reform, and sensitivity to country characteristics (World Bank, 213c). For example, Jordan s MTEF was a component of major public financial management reforms in 24 and part of the national development strategy. The MTEF s specific objective was to improve fiscal discipline through realistic revenue projections, followed by better expenditure prioritization and the identification of fiscal space. In the case of South Africa, the MTEF was introduced in the context of high government debt and a combination of underspending by the central government and overspending by provincial governments. Underspending and overspending were both reduced following the introduction of the MTEF. One of the lessons from the experiences of South Africa, Tanzania, and Uganda is the need for realistic expectations during the preparation of the budget, without which even well-designed MTEFs cannot succeed (Holmes and Evans, 23). Risks and Medium-Term Objectives While debt stocks in many developing economies remain moderate, primary deficits are wider than they were before the crisis. Although debt has grown slowly under the current benign market conditions, especially low interest rates, the debt-to-gdp ratios could increase much more rapidly if domestic growth slows and global interest rates rise (Figure 27For example, Le Houerou and Taliercio (22) examine the design and implementation of MTEFs in a sample of African economies. 135

18 FIGURE 3.13 Sustainability gaps under different conditions in 213 In some EMEs and FMEs, fiscal risks would increase under historic market conditions. A. Current market conditions B. Historic market conditions Note: The sustainability gap is the difference between the primary balance and an estimated debt-stabilizing primary balance, which depends on assumptions about interest rates and growth rates. For a given country, current market conditions refer to 213 interest and growth rates, while historic conditions refer to the sample average during A negative value suggests that the balance is debt-increasing, a value of zero suggests that the balance holds debt constant, and positive values suggest that the balance is debt-reducing. A redder color indicates a more negative sustainability gap; a greener color a more positive gap. If the data was updated to 214, some countries would show more benign sustainability gaps (e.g. Spain) while others would show lower ones. 136